Opus 5.5 is 20% cheaper per input and output token than Opus 5, and 60% cheaper on cache reads. So what does that actually do to the cost of a task in Claude Code? We ran the numbers, and built a calculator so you can run yours from /usage: claude.dev/blog/what-a-task-…

Sep 25, 2026 · 6:13 PM UTC

251
264
6,202
739,810
Sort replies: Relevant Recent Liked
Replying to @ClaudeDevs
It's great so far please do not nerf or fuck this up.
5
248
7,886
Replying to @ClaudeDevs
Great calculator for understanding real world coding costs
700
Replying to @ClaudeDevs
We talking ultracode, Max…?
269
Replying to @ClaudeDevs
1
95
11,920
Replying to @ClaudeDevs
Don't you guys dare nerf it.
27
2,271
Replying to @ClaudeDevs
I love Opus 5.5
How a WhatsApp message travels in under a second. Built it and animated it with Opus 5.5.
1
3
15
3,572
Replying to @ClaudeDevs
please dont
"X sentiment scores for Opus 5.5 are above 0.9. Nerf the model."
1
22
5,935
Replying to @ClaudeDevs
?
1
19
2,926
Replying to @ClaudeDevs
All this is fine just don’t nerf it up. No kidding. Don’t nerf it up. I repeat don’t nerf it up.
18
1,587
Replying to @ClaudeDevs
This is the first model ever that I have absolutely no complaints or even suggestions for. Incredibly amazing. Just don't nerf it up please.
13
1,437
Replying to @ClaudeDevs
Yo, we all love Opus 5.5 This release will be marked in history… and so will the day if you choose to nerf it
1
11
2,133
Replying to @ClaudeDevs
I've been a hater but have to admit y'all cooked with Opus 5.5 Upgrading my Claude sub & downgrading my gpt sub Keep cooking and I'll end up getting rid of my openai sub and I'll be fully on team Anthropic/Spacex
13
1,144
Replying to @ClaudeDevs
Thanks for sharing the great article. Especially the interactive elements of the article was really helpful to visualize the comparative analysis and how it could affect the usage and the results of using Opus 5.5. Keep it up.
5
5,822
Replying to @ClaudeDevs
So you can release a better yet cheaper model The only thing needed was competition
2
3,780
Replying to @ClaudeDevs
I can see the difference which is pretty noticeable!
3
3,458
Replying to @ClaudeDevs
Incredible work with 5.5! Please don’t change a thing with it 🙏🏻
2
654
Replying to @ClaudeDevs
I don't need a calculator to know the answer: It costs 0. 😅 (at least that's what if feels right now - PLEEEAASE don't ever change that!! 🙏)
2
827
Replying to @ClaudeDevs
the calculator showing me i've been spending way more on input tokens than i thought
1
1,204
Replying to @ClaudeDevs
who needs a calculator, run the task twice and you already know which one's cheaper
1
4,799
Replying to @ClaudeDevs
@grok what does this article say re subscription accounts? What’s the rough ‘extra usage’ subscription accounts can expect to see, if any?
1
1
11,541
Replying to @ClaudeDevs
I think this understates how much cheaper 5.5 is over 5. It seems more token efficient on top of being cheaper per token.
3
1,859
Replying to @ClaudeDevs
Whatever you guys have done is the best thing after Opus 4.8 and don't nerf it at all please!
1
602
Replying to @ClaudeDevs
Ficou tão mais barato que agora meu plano acabou dias antes do reset pela primeira vez. Se continuar assim, foi dizer adeus e vou pro Codex.
2
1
1,050
Replying to @ClaudeDevs
So when are guys gonna nerf opus 5.5 just like fable?
3
1,446
Replying to @ClaudeDevs
20 percent cheaper tokens and 100 percent more expensive excuses for not shipping
1
634
Replying to @ClaudeDevs
What you guys did with Opus 5.5 is amazing, much more efficient, and costs reduced about 30% for me, maybe more, since this week I used it more than usually did with Opus 5. The shorter and clearer responses are one of the best upgrades comparing to Opus 5. Nice job!
14
Replying to @ClaudeDevs
20% cheaper, 200% chattier — the cow gives less milk and more moo, and you pay for both.
1,047
Replying to @ClaudeDevs
It's nurfed already. Stay away from Anthropic.
1,273
Replying to @ClaudeDevs
It was amazing the first few days when it was released, now its not the same at all, they did some nerfing for sure...
1
1,337
Replying to @ClaudeDevs
🙏🙏🙏
Not hitting usage limits no matter what I do is the biggest gift I've ever received on a 20$ AI plan. PLEEEAASE claude, don't ever take that away from me again!! 🙏🥹
854
Replying to @ClaudeDevs
Hey, can you tell me why my session usage is running out quickly? I am using a guest pass. But afaik guest pass doesnt have lower limits right? I want to know if theres something wrong with my account or the limits are nerfed on newer accounts before I fully commit
1
4,157
Replying to @ClaudeDevs
This is so useful—a calculator straight from /usage!
1,558
Replying to @ClaudeDevs
Just switched from codex. Absolutely amazing. Can’t wait to switch back to codex when you inevitably enshittify Opus 5.5
2
1,179
Replying to @ClaudeDevs
Oh so THAT's why I've barely touched my weekly limit since 5.5
1
179
Replying to @ClaudeDevs
60% cheaper on cache??
1,401
Replying to @ClaudeDevs
Now , no one will stop me from building my own GTA- 6
20